OpenAI and Upwork Study Finds AI Freelance Projects Pay More

AI-related freelance projects on Upwork carried larger fixed-price budgets and involved more hours of work than comparable non-AI projects, according to new research released jointly by OpenAI and Upwork. The study, part of OpenAI's "Work at the Frontier" series, analyzed 63 work categories across Upwork job postings and direct activity in ChatGPT between early 2023 and late 2025.

The findings do not support a simple story of AI replacing freelance labor. Instead, the researchers found shifting demand for familiar services, new combinations of skills bundled into single projects, and expanded budgets for work that incorporates AI capabilities.

The economic stakes are large. Upwork's 2026 Future Workforce Index found that 38% of skilled U.S. knowledge workers freelance. A previous annual study by the company estimated that freelance knowledge work generated $1.5 trillion in U.S. earnings in 2024. Those figures extend well beyond Upwork's own marketplace.

Why freelance data matters for tracking AI's effects

Changes in the freelance market arrive one assignment at a time, the researchers noted: a client needs a different service, requests a new skill, or expands a project's scope. Freelance contracts are often short, and clients can adjust their requirements with each posting.

That structure makes freelance postings a faster signal than economy-wide employment statistics. "Freelance work can also reveal changes sooner than economy-wide employment statistics," the OpenAI Economic Research Team wrote in the report. "Contracts are often short, and clients can adjust their requirements with each posting."

Work inside firms can evolve without ever appearing in a new job posting. Freelance marketplaces publish those changes continuously. Combined with ChatGPT usage data, the postings show which activities people now perform directly with AI, which they still hire freelancers to do, and how AI-related projects differ from other work.

What did the 63-category comparison show?

The study compared task activity across two platforms, and the mix of work differed sharply between them. Writing accounted for a much larger share of ChatGPT use, while video and animation accounted for larger shares of requests in Upwork postings.

Across the 63 work categories analyzed, 43 followed a consistent pattern: activities that were relatively common in ChatGPT lost share on Upwork, while less common activities gained share. The remaining 20 categories showed the reverse pattern.

The researchers were explicit about the limits of that correlation. ChatGPT use alone does not explain changes in freelance demand, they wrote. The comparison also captures only how people currently use ChatGPT, not the full range of tasks AI can support — a gap OpenAI calls the "capability overhang."

How are AI-related job postings changing?

By late 2025, Upwork clients were requesting AI in many more, and different kinds of, work. Freelancers were sought to apply AI in tasks including:

• Chatbot development
• Video production
• Marketing automation
• Application integration

Within individual projects, AI capabilities were frequently requested alongside other work. AI-related postings listed roughly as many tasks as other postings, but more often combined tasks that had rarely appeared together in earlier jobs.

One example from the report: a single posting requested AI-generated video, video editing, and motion graphics, with the freelancer delivering a finished video by bringing all three skills together.

How much more diverse did AI work become?

Between early 2023 and late 2025, the mix of tasks in AI-related postings became 45% more diverse. The researchers quantified this as the equivalent of an increase from about 15 equally common tasks to about 21.

• Early 2023: 14.6 equally common tasks
• Late 2025: 21.2 equally common tasks

This diversification is the study's clearest structural signal. Clients are not simply substituting AI for existing freelance services; they are assembling projects from broader and less conventional combinations of skills.

What about rates and budgets?

Posted hourly rates were similar for AI-related and other work, the study found. The difference appeared in fixed-price budgets and total hours: AI-related projects had larger fixed-price budgets and involved more hours of work.

The researchers offered a measured interpretation of that gap. The pattern suggests that higher budgets may partly reflect a broader scope of work rather than premium pricing for AI skills alone.

What are the limits of the findings?

The OpenAI Economic Research Team framed the study as an early view, not a verdict. As the group has noted in previous "Work at the Frontier" papers, AI's effects on work can be difficult to detect in employment statistics, and the effects may appear first in the work people do — not in aggregate job counts.

The study's core caution bears repeating: the data captures current ChatGPT usage, not the full set of tasks AI could support. OpenAI's own term for that gap, the capability overhang, implies that freelance demand may shift again as users discover more of what the technology can do.

What comes next?

The findings point in two directions at once. Some activities common in ChatGPT use lost share on Upwork, but the pattern was not uniform across categories. AI-related work spanned a wider mix of activities, combined tasks in less familiar ways, and carried larger fixed-price budgets and more hours.

"Together, these patterns suggest that AI may be shifting both what clients seek from freelancers and how they assemble projects," the researchers concluded.

The OpenAI Economic Research Team says it will continue tracking how AI changes work in real time through the Work at the Frontier series. For freelancers, the near-term signal is concrete: budgets and hours are growing on AI-related projects, and the skill combinations clients request are broader now than at any point since early 2023 — a shift measurable at 45% in just under three years.